The SiteTool is an R Shiny tool used to assess the land cover characteristics of potential field sites.
The aim of this tool is to help researchers select sites in a quantitative way using spatial data. Often, researchers aim to have field sites that exist along a gradient of different land use types and characteristics. This tool allows researchers to analyze the land cover characteristics of a list of sites, and see where potential sites fall along a gradient of other potential sites.
This tool is also available as a local R package https://github.com/BioDivHealth/sitetool, which allows larger file uploads.
The tool is broken down into three main steps. It is designed so that you can go back and forth between steps to update your site list.
There are three ways to select an area of interest:
Box Tool
Select the box icon located in the upper left-hand corner of the map. Use this to select your potential study site area.
Bounding Box Coordinates
You may also manually enter the coordinates of your bounding box. The map will update based on these bounding box coordinates.
Shapefile
You can upload a shapefile covering your area of interest. Please ensure shapefiles are in GeoJSON format. The shapefile will be displayed once uploaded. A good tool for creating a shapefile for your area of interest in https://geojson.io/.
Once you have selected an area of interest, you will need to add raster data related to your selection criteria.
Default Raster Data
Use the drop down menu to select a type of raster data. There are three default raster sources: ESA WorldCover, SRTM Elevation, and Human Footprint data. Once you have selected a source press Add Raster. The raster will be displayed on the map when finished.
Upload your own data
Select Upload a file from the drop down menu. Press Browse and navigate to the raster on your computer. Uploaded rasters must be in GeoTiff format. If your area encompasses multiple tiles, please merge before uploading. If you are using a categorical raster, make sure there is a color table associated with it for correct display. If you are using the browser version, please ensure you are using Google Chrome, as it is the only browser that supports GeoTiffs. (Note: the browser version only accepts uploads less than 5 MB, while the local version can accept uploads up to 100 MB).
You can continue adding rasters to cover all of your parameters of interests. Please use the map toggle betweeen visualizing the different data sources.
You must select a sampling procedure in order to generate a list of sites. You may also skip this step by leaving sampling procedure as None and only using selected sites (see instructions below).
Random Sites
Sites will be selected based on simple random sampling. Only land areas will be selected, and bodies of water will be avoided. You must specify the number of sites needed and the distance between the sites. You can also specify that sites must be a certain distance from a major city and/or major roads.
Village Sites
Village sites will find all of the towns/villages/cities in the selected area using OpenStreetMap. You can have the tool find all of the villages in an area or randomly select a subset from that region. Note that if you select a very large area for village sites, the request may time out.
These are your “shortlist” sites that you are considering for the study. If you don’t have any selected sites yet, you can leave this option as None. Otherwise, there are two options for adding selected sites:
Select on Map
Select this option to toggle a tool that allows you to click on the map to add sites to the selected list. Clicked areas will show up as red points. If you need to clear the points, please select the option below the map.
Upload a CSV
Select this option to upload a list of sites in CSV format. The CSV should contain a column labeled site, which has the site name, and columns with latitude and longitude in decimal degrees.
In this step, you will analyze the land cover data around each site from your uploaded rasters.
Select analysis distance
Input the distance from the center of each potential site you would like the land cover data analyzed from in meters. This distance is the distance from the point to each edge of the raster on all four sides, so a distances of 1000 meters (1km) would lead to an area of analysis of 4 km2.
Press Go. The calculations may take a while to run. The plots and data will be displayed in the respective tabs when finished.
Summary Plot
The plot shows the values of your input sites relative to the other sites analyzed. If you hover over a point, it will show the value for that site across of all over the analyzed raster data. . For categorical rasters, the proportion of that land cover type found within the analysis distance is shown. For continuous rasters, the mean value of that measurement is shown within the analysis distance.
You can add and remove sites from your “generated” and “selected
lists. The map will also update. In this example, all snowy sites were
added to the selected sites.
Statistical Comparison
The stats tab indicates whether the selected sites are statistically different from the generated sites across land cover values.
Data Tables and Export
The data tables contain the results of the analysis for each site and can be exported. The following measurement values for categorical rasters are included:
For continuous rasters, the values calculated are mean, min, and max for the analysis area.
Press the Save Dataset at the top of the table to export the data. You export just the selected sites or the full dataset depending on which tab is selected.
The app can support categorical land cover rasters such as:
This repository has a series of Google Earth Engine scripts to assist with exporting this data: https://github.com/BioDivHealth/GEE_Scripts. You can also visit the above sites to download a raster for your region of interest. If your region covers multiple tiles, please merge tiles into one file prior to uploading. Please ensure the raster CRS is WGS 84, and the values for any land cover type are in one layer following the numeric codes for the above products. If your raster has a defined CRS but is not in WGS84, the app will reproject it to WGS84 (EPSG:4326) before analysis.
The app can also support rasters with continuous values, such as for temperature, NDVI, or elevation. For these inputs, please ensure the input raster has only one numeric layer.
It is recommended to use high-resolution products (< 100m) due to the scale of the analysis.
If you have any questions, encounter a problem, or would like to suggest improvements: